MEMORY
FOR
LOCATIONS
TITS
OF STORED
AND MARSH
FOOD
IN WILLOW
TITS
by
SUSAN D. HEALY1,2) and JUKKA SUHONEN3,4)
(1Department of Psychology, University of Newcastle upon Tyne, Newcastle upon Tyne
NE1 7 RU, UK; 3Department of Biological and Environmental Sciences, PO Box 35,
FIN-40351 Jyväsklyä, Finland)
(Acc. 14-VI-1995)
Summary
Some food-storing animals use spatial memory in food retrieval. There is variation amongst
food-storing species in the duration over which food is stored and it is positively correlated
with differences in the size of the part of the brain that is involved with memory for storage
locations, the hippocampus. For example, British marsh tits (Parus palustris) are generally
thought to retrieve their stores after a few hours or days whereas Scandinavian willow tits
(P. montanus) have been shown to store their food for several weeks to months. Marsh tits
also have a smaller hippocampus than do willow tits. This experiment tested the hypothesis
that retrieval success (and thus memory capacity) of the two species differs depending on
the usual length of time for which food is stored. Individuals of both species were tested
on their ability to retrieve five stored food items after two differing delays: 2-3 h ('short')
and 17 days ('long'). The prediction was that there would be no differences in retrieval
performance between the species after the short delay but that willow tits would be more
accurate at retrieving their stores than marsh tits after the long delay. The results did not
provide support for this hypothesis as there were no differences between the species in
retrieval performance following either delay: both species successfully retrieved food after
both delays.
2) e-mail address: [email protected]
4) We thank Alan Woodington for catching the marsh tits. Thanks also to Anders Brodin,
Dan Cristol, John Krebs, Tim Guilford, three anonymous reviewers who gave valuable
advice and useful comments on the manuscript and to Jonathan Newman for his tireless
statistical advice. St John's College, Oxford, the Royal Society and the Academy of Finland
provided financial support.
72
Introduction
in a taxonomically
Food storing has been documented
wide range of spefrom
ants
and
bees
to
bears
and
cies,
monkeys (reviewed in Vander Wall,
1990). There is an accumulating body of evidence that animals use memory
to retrieve their stores (Sherry, 1989; Vander Wall, 1990) and that this
and large. For example, food-storing
birds in
memory is multi-faceted
the families
Paridae
and Corvidae
may store many food items, e.g. Clark's
nutcrackers (Nucifraga coluinbiana) may store as many as 33 000 hoards in
an autumn (Tomback, 1978; Vander Wall, 1983). They appear to remember
the quality of the food they store in each site (Sherry, 1984a) and on
retrieval seem to recognise that they have emptied the site and avoid it
& Krebs, 1982; Sherry, 1984b).
(Shettleworth
subsequently
There is variation in storing strategy both within and among
species.
ways in which the strategies vary is in the number of food items placed at each storage site. For example, scrub jays
will often put several seeds into a storage site
(Apheloconia coerulescens)
whereas another storing corvid, the magpie (Pica pica) usually stores items
One of the most obvious
singly. The length of time over which food is stored also varies. Clark's
nutcrackers, for example, relocate stored seeds after as long as 1 1 months
(Vander Wall, 1983), whereas marsh tits (P. palustris) usually retrieve their
a few hours or days (Cowie et al., 1981; Stevens & Krebs,
variation
This
1986).
provides an opportunity to examine whether there is
a specificity to the memory capacity involved in food retrieval. For examstores
within
ple, do those species which store food for shorter periods of time forget
storage sites more quickly than those species which retrieve their food after
much longer intervals?
Available field data suggest that variation in storing strategy between
the four storing species of north-western
European Paridae is very much
less than it is within
the Corvidae.
The different tit species tend to use
the same kind of storage sites, store singly, similar insect and plant seed
species (Haftom, 1956a, b; Alatalo & Lundberg, 1983) and space storage
sites at similar distances (Sherry et al., 1982; Brodin, 1992). The most
apparent differences between the species in storage are the length of time
which elapses before food is retrieved and the number of items stored.
species which store tens or hundreds of food items and retrieve
their stores within a few days or a few weeks include coal tits (P. ater)
Those
73
1956a), marsh tits (P. palustris)
(Stevens & Krebs, 1986), and
chickadees
(P. caeruleus) (Sherry, 1989). Species which store
black-capped
thousands
of seeds and do not retrieve their stores for several months
(Haftom,
include
the willow
tit (P. montanus).
Observations
document individuals
of this species retrieving Galeopsis seed in December-March,
seeds which
must been stored some three to four months earlier (Haftom, 1956b).
The behavioural variation in food storing is accompanied by differences
in the part of the brain involved in processing memories for storage sites,
the hippocampus
1966; Sherry & Vaccarino, 1989): food(Krushinskaya,
have
a
than non-storers (Krebs et al.,
storing species
larger hippocampus
1989; Sherry et al., 1989). This difference is dependent on the degree of
food storing: those species which store more and for longer have a larger
hippocampus
(Healy & Krebs, 1992a, 1995). In addition, developmental
studies have shown that the enlargement of the hippocampus occurs at the
time food storing begins (Healy & Krebs, 1993; Clayton & Krebs, 1994;
Healy et al., 1994).
One of the possible advantages of an enlarged hippocampus is a greater
In the experiment presented here we
ability to retain spatial information.
tested the hypothesis that marsh tits, which store food for relatively short
intervals (hours), will not be as able to remember multiple food storage
sites as willow tits, which have a larger hippocampus
than marsh tits and
store food for weeks to months. The retrieval performance of wild-caught
individuals of these two species was compared after a 'short' delay (2-3 h)
and a 'long' delay (17 days). The short delay was chosen as representative
of the time period over which marsh tits are known to retrieve. The longest
time period over which marsh tits have been seen to retrieve stores is about
three days, which is well short of the 17-day period. We did not want to
test the birds over delays typical of willow tit storage time (30-90 days)
as we could not be sure that the willow tits, in a laboratory setting, would
perform well. We predicted, therefore, that both species would accurately
retrieve stored seeds after the short interval, whereas the willow tits would
perform better than the marsh tits after 17 days.
Subjects
The birds used in this study were seven wild-caught willow tits from coniferous woodland
near Konneversi, Finland and nine wild-caught marsh tits from deciduous woodland near
74
Oxford, England. Both species were maintained on a diet of insectivorous bird mix, peanuts,
and sunflower seeds, with fresh water provided daily. The birds were kept indoors in
individual cages (0.44 m wide x 0.77 m long x 0.44 m high) that were connected to the
experimental room by electronic doors. The experiment was run in February-March 1991
and November 1991-February 1992 at the Department of Zoology at Oxford and birds were
maintained on an 8 : 16 light: dark cycle.
Experimental environment
The birds were tested in a room measuring 3.75 m x 3.9 m x 2.4 m high. Along the 3.9 m
wall were twelve black electronic doors through which the birds flew into the room and the
3.75 m wall had a door with a one-way Plexiglas window through which the experimenter
could view the birds. In the experimental room were six 'trees' which were tree branches
set into white plastic umbrella stands, arranged in approximately circular fashion. On each
of the trees were 10 potential storage sites: small holes (0.5 cm diameter, 1-cm deep) drilled
into the trees no less than 15 cm apart (providing a total of 60 potential storage sites). The
holes were placed irregularly on the tree's trunk and branches. Each hole was covered with
a piece of black cloth (2.5 x 2.5 cm) stapled just above, and covering, the hole such that
the bird was unable to see the contents of the hole without raising the cloth. This cloth was
readily raised by all the birds, both when storing and retrieving.
At the centre of the ring of storage trees were two bowls of food in which were storable
food items: one contained pieces of peanut and husked sunflower seeds, the other wax moth
larvae. Wax moth larvae were only ever available to the birds in this experimental room.
Experimental procedure
Throughout the experiment the birds' food bowls were removed at dark (16:30 GMT). The
birds were tested between 09:00 and 14:00 and were subsequently fed a fresh portion of
the maintenance diet.
The birds were trained to enter the room when the door connecting their cage to the
experimental room was opened and then to fly to the food bowls. They were then allowed
to eat from either bowl and storing usually ensued. Over the course of five to seven days
the birds were trained to store only in the storage sites on the experimental trees. This is
done by consistently removing all items stored in 'illegal' sites. During the experiment the
bird was allowed to store only five items. When this had occurred the bird's electronic door
was opened and the lights in the experimental room turned off.
In the retrieval phase of the experiment the bird was allowed to return to the experimental
room after a retention interval in order to retrieve its stored food. The food bowls had been
removed and the contents of each hole were covered by the cloth flaps. For each bird food
was replaced in the five holes in which it had stored previously. The bird was observed
until it had visited a maximum of 10 sites. The retrieval phase typically lasted for about five
minutes. In this experiment there were two delays at which each bird's retrieval performance
was tested. There was a 'short' delay of 2-3 h and a 'long' delay of 17 days. On short-delay
75
days the birds were not fed during the retention interval. Birds were not deprived of food
nor did they go in to the experimental room during the seventeen day interval.
When the birds had learned to store and retrieve in the experimental room they were
given five short-delay tests followed by a long-delay test. The data recorded in the retrieval
phase were the ten sites which birds inspected. Revisits within a retrieval phase were very
uncommon and were not recorded.
Results
ability of the birds was assessed by determining how many
stored items they found in the retrieval phase. For the short delay the mean
number of successful retrievals (seeds found within the 10 visits allowed
The retrieval
during the retrieval phase) made over the five days was taken as the bird's
score. For the long delay the bird's score was taken as the number of
successful retrievals made in the first 10 visits of the retrieval phase.
tits took significantly longer than willow tits to store five items
0.73 min (meant
0.98 min; willow tits: 9.87
SE);
(marsh
=
U-test: U = 10.5, U'
52.7, NI = 7, N2 = 9, p = 0.026).
Mann-Whitney
There was no difference between the species in time taken to retrieve stores
Marsh
tits:
(marsh
tits:
Mann-Whitney
Performance
0.46 min; willow tits:
U = 14, U' = 49, Nl =
U-test:
with respect
0.45 min (mean + SE);
7, N2 = 9, p = 0.064).
to random
were performing
on retrieval with
respect to random it is first necessary to show whether the birds were
If the distribution
of storage sites is no
using storage sites at random.
different from that of a random site usage then the calculation of retrieval
In order
to determine
how the birds
can be based on all 60 potential storage sites in the room.
performance
If the birds were using sites for storage differently from that of a random
expectation then analysis of retrieval cannot include sites birds would not
was used to calculate, for five successive
visit. The binomial distribution
that the birds used unique sites on each day. These
days, the probabilites
were then used to generate the expected number of unique
probabilities
storage sites used over 5 days. A Z-test showed no significant difference
storage sites used by
the birds over the 5 days of the short delay (average no. of sites used =
19.68 ::L 0.68 SE, N = 16) and the expected number of unique sites used
between
the distribution
of the number
of unique
76
over 5 days (expected number =
19.84; z = -0.66, p = 0.51, two-tailed
This
result
that
the
birds did not have marked preferences
test).
suggests
for particular sites and therefore we assumed that the birds were effectively
choosing storage sites at random.
Z-tests were used to test whether
the birds' retrieval success was better than that expected by chance. A hypergeometric
distribution was used
to calculate the expected number of successes (0.83). The retrieval performance of both species was better than random after the short delay of
2-3 h (willow tits: z = 7.47, p < 0.0001, N = 7; marsh tits: z = 8.26,
p < 0.0001, N = 9). Both species also performed better than at random
= 7.70, p < 0.0001, N = 7; marsh tits:
after 17 days (willow tits: z
z = 7.5, p< 0.0001, To = 9).
There was no difference between the species at either delay (2-3 h:
30, U' = 33, Ni = 7, NZ = 9, p = 0.87; 17 days:
Mann-Whitney U =
U = 27.5, U' = 35.5, Nj = 7, N2 #
9, p = 0.64) (see Fig. 1).
Fig. 1. The mean number (t SE) of food items found by marsh tits (N = 9) and willow
tits (N = 7) after retention intervals of 2-3 h and 17 days. The solid line indicates how
many seeds would be retrieved if the bird retrieved its stored food at random.
77
species was there a difference
delays (Wilcoxon paired sign tests: willow
tits: p = 0.40, N = 9). This result could be
all birds were tested with short delays before
For neither
in performance
between the
tits: p = 1.0, N = 7; marsh
due to an effect of training as
the long delay.
Discussion
are (i) marsh tits and willow tits
The main results from this experiment
did not differ in their ability to retrieve their stored food; (ii) both species
performed better than at random after both short and long delays; (iii) after
17 days neither species' retrieval performance was significantly worse than
it was after 2-3 h (Fig. 1).
This experiment was designed to test the hypothesis that a bird's ability
to remember the locations of stored
(as measured by retrieval performance)
needs to be
with the length of time the information
food is correlated
for. It was proposed that willow tits, which store their food for
several weeks to a few months, would show high retrieval accuracy after a
long delay. As marsh tits store food for only a few hours or days it would
seem unnecessary for them to remember storage locations for long periods
retained
and they were not expected to perform well after a long delay. The data
from this experiment show that the marsh tits are as capable of successful
retrieval for at least 17 days, even though this is considerably longer than
the time period over which they have been observed to store (Cowie et al.,
1981; Stevens & Krebs, 1986).
for the failure to find a close
There are several possible explanations
coupling between retrieval success and memory demand in this experiment.
The first possibility is that the general belief that willow tits store food for
Thus it may be that marsh
longer than do marsh tits may be inaccurate.
tits store food for longer than has been observed so far. It is possible that
even though the birds may recover most of their stores within a few hours
or days, some part of their store is left for retrieval at some much later
date.
A second possibility is that storage locations are remembered until food
of time. Whilst the passage of time
is retrieved from them, independent
alone does not explain forgetting (see Baddeley, 1990) most theories would
predict that storage locations would over lengthening periods of time be
78
remembered
derwood,
less well, due in part to the effects of interference (e.g. Un1957; Squire, 1987). For the marsh tit in the field storing and
on a daily basis, there must be competition,
to some extent at
retrieving
least, between memories
even higher
presumably
for different
Interference
is
storage locations.
because the birds appear to remember emptied
sites
order
to
avoid
(in
them) in addition to baited sites. However,
storage
in this experiment the lack of food storing in the experimental room during
the retention period may have decreased interference,
and thus memory
loss, especially given the small number of storage sites to be remembered.
Good performance
by the marsh tits, then, after 17 days may have been
achieved by low levels of interference and therefore their performance
in
this experiment may not be fully indicative of memory abilities in the field.
The ability of these birds to retrieve stored food after 17 days as well
as after 2-3 h is impressive when considered in the context of the results
from other laboratory studies of animal memory.
Rats and pigeons, for
both
a
marked
show
decline
in
within a
example,
memory performance
few hours (e.g. Roitblat & Harley, 1988; Spetch & Honig, 1988). However, the nature of the tasks these species are tested on is undoubtedly
the single largest factor in explaining such apparently poor performance.
tasks show
Food-storing
species tested on delayed-non-matching-to-sample
memory deficits after a few minutes (Olson, 1991; Healy & Krebs, 1992b;
Brodbeck & Shettleworth,
1995) whereas when tested on retrieval of food
these
birds
stores,
may perform better than expected by chance up to at
least 28 days (Balda & Kamil, 1989; Hitchcock & Sherry, 1990).
The lack of difference in performance
between the two species
may
in
speculation as to the role of the hippocampus
birds. As in this comparison of two storing species, those testing memory
retention over differing periods of time (30 min-24 h) between non-storing
and closely related storing parids have not shown substantial differences
between the species (great tits and marsh tits: Healy & Krebs, 1992b;
contribute
to the current
blue tits and marsh
Healy & Krebs, 1992c), even though there are
considerable
differences
in hippocampus
size. Those results led to the
that
the
differences
in
size may be more closely
suggestion
hippocampus
correlated with the amount of information stored rather than the duration
over which memories are kept and the results from this experiment lend
support
tits:
to this hypothesis.
79
References
Alatalo, R.V. & Lundberg, A. (1983). Laboratory experiments on habitat separation and
foraging efficiency in marsh and willow tits. - Ornis Scand. 14, p. 115-122.
Baddeley, A. (1990). Human memory: Theory and practice. - Lawrence Erlbaum Associates Ltd., Hove.
Balda, R.P. & Kamil, A.C. (1989). A comparative study of cache recovery by three corvid
species. - Anim. Behav. 38, p. 486-495.
Brodbeck, D.R. & Shettleworth, S.J. (1995). Matching location and color of a compound
stimulus: Comparison of a food-storing species and a non-storing bird species. J. Exp. Psych.: Anim. Beh. Proc. 21, p. 64-77.
Brodin, A. (1992). Cache dispersion affects retrieval time in hoarding willow tits. - Ornis
Scand. 23, p. 7-12.
Clayton, N.S. & Krebs, J.R. (1994). Hippocampal growth and attrition in birds affected by
experience. - Proc. Natl. Acad. Sci. USA 91, p. 7410-7414.
Cowie, R.J., Krebs, J.R. & Sherry, D.F. (1981). Food storing by marsh tits. - Anim. Behav.
29, p. 1252-1259.
Haftorn, S. (1956a). Contribution to the food biology of tits especially about storing of
surplus food. Part II. The coal-tit (Parus a. ater L.). - Det Kgl. Norske Vidensk.
Selsk. Skr. 2.
-(1956b). Contribution to the food biology of tits especially about storing of surplus
food. Part III. The willow-tit (Parus atricapillus L.). - Det Kgl. Norske Vidensk.
Selsk. Skr. 3.
Healy, S.D., Clayton, N.S. & Krebs, J.R. (1994). Development of hippocampal specialisation
in two species of tit (Parus spp.). - Behav. Br. Res. 61, p. 23-28.
-& Krebs, J.R. (1992a). Food storing and the hippocampus in corvids: amount and
size are correlated. - Proc. R. Soc. 248, p. 241-245.
-&-(1992b). Delayed-matching-to-sample by marsh tits and great tits. - Q. J.
Exp. Psych. 45B, p. 33-47.
-&-(1992c). Comparing spatial memory in tits: recalling a single positive
location. - Anim. Learn. Beh. 20, p. 121-126.
-- & -(1993). Development of hippocampal specialisation in a food-storing bird.
- Behav. Brain Res. 53, p. 127-131.
-- & -(1995). Food storing and the hippocampus in Paridae. - Brain Behav. Evol.
Hitchcock, C.L. & Sherry, D.F. (1990). Long-term memory for cache sites in the blackcapped chickadee. - Anim. Behav. 40, p. 701-712.
Krebs, J.R., Sherry, D.F., Healy, S.D., Perry, V.H. & Vaccarino, A.L. (1989). Hippocampal
specialization of food-storing birds. - Proc. Natl. Acad. Sci. USA 86, p. 1388-1392.
Krushinskaya, N.L. (1966). Some complex forms of feeding behaviour of nutcracker Nucifraga caryocatactes, after removal of old cortex. - Zh. Evol. Biochim. Fisio. 11,
p. 563-568.
Olson, D. (1991). Species differences in spatial memory among Clark's nutcrackers, pigeons
and scrub jays. - J. Exp. Psych.: Anim. Beh. Proc. 17, p. 363-376.
Roitblat, H.L. & Harley, H.E. (1988). Spatial delayed matching-to-sample performance by
rats: learning, memory and proactive interference. - J. Exp. Psych.: Anim. Beh.
Proc. 14, p. 71-82.
80
Sherry, D.F. (1984a). Food storage by black-capped chickadees: memory for the location
and contents of caches. - Anim. Behav. 32, p. 451-464.
-(1984b). What food-storing birds remember. - Can. J. Psychol. 38, p. 304-321.
-(1989). Food storing in the Paridae. - Wilson Bull. 101, p. 289-304.
--,
Avery, M. & Stevens, T.A. (1982). The spacing of stored food by marsh tits. Z. Tierpsychol. 58, p. 153-162.
-& Vaccarino, A.L. (1989). Hippocampus and memory for food caches in black-capped
chickadees. - Behav. Neurosci. 103, p. 308-318.
--, --,
Buckenham, K. & Herz, R.S. (1989). The hippocampal complex of foodstoring birds. - Brain Behav. Evol. 34, p. 308-317.
Shettleworth, S.J. & Krebs, J.R. (1982). How marsh tits find their hoards: The role of site
preference and spatial memory. - J. Exp. Psych.: Anim. Beh. Proc. 8, p. 354-375.
Spetch, M.L. & Honig, W.K. (1988). Characteristics of pigeons' spatial working memory
in an open-field task. - Anim. Learn. Behav. 16, p. 123-131.
Squire, L.R. (1987). Memory and brain. - Oxford University Press, New York.
Stevens, T.A. & Krebs, J.R. (1986). Retrieval of stored seeds by marsh tits Parus palurtris
in the field. - Ibis 128, p. 513-525.
Tomback, D.F. (1978). Foraging strategies of Clark's nutcrackers. - Living Bird 16, p. 123161.
Underwood, B.J. ( 1957). Interference and forgetting. - Psychol. Rev. 64, p. 49-60.
Vander Wall, S.B. (1983). Remembrances of seeds stashed. - Nat. Hist. 92, p. 60-65.
-(1990). Food hoarding in animals. - University of Chicago Press, Chicago.